OpenAI Launches Full-Duplex GPT-Live Voice Models
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI launched GPT-Live, a new generation of voice models built on a full-duplex architecture that lets them listen and speak simultaneously, now powering ChatGPT Voice.
- GPT-Live-1 and GPT-Live-1 mini are the two models introduced, with GPT-Live-1 rolling out in ChatGPT immediately and API access for both opening via a developer signup form.
- Sam Altman said GPT-Live "feels magical and 'real'" and predicted it could shift users from typing to talking — a notable reversal from someone who previously preferred typing.
- Greg Brockman said OpenAI is working on bringing GPT-Live to the API and Codex, signaling plans to extend the full-duplex capability to developer tools and coding workflows.
- Coverage from Engadget and The Verge highlighted that the new voice mode will slow down when asked and is "better at shutting up," surfacing real-time behavioral control and improved turn-taking as wins that go beyond the "more natural" framing.
- The launch drew widespread coverage across CNET, TechCrunch, 9to5Mac, TechRadar, The Mac Observer, and RuntimeWire, with early testers describing the latency drop and conversational feel as a "step change."
Why it matters: Altman — who previously preferred typing — publicly says GPT-Live could change how users interact with AI, and OpenAI is already telegraphing API and Codex rollouts that would put the full-duplex capability in developers' hands for customer service, agents, and coding tools.
